suspicious - Story 145598 | Emotion Vectors in Gemma4-31B

The rejection letter from "Golden Dawn Press" landed on Emily's desk like a dead bird. She felt her stomach churn. The language was carefully chosen, yet the sentiment felt… malicious. It was overly courteous.

She reread the letter, each word digging into her. The rejection was vague, offering no real reason for the decision. She'd spent months perfecting her manuscript. She felt a chill crawl up her spine.

She felt the need to get to the bottom of this. She logged onto the internet, then typed the editor's name, "Arthur Finch." It was then that she discovered his blog, which he had not updated in years. It was then that she found a single post. A veiled and cruel post, about a writer who had betrayed him.

Emotion: suspicious

Cluster: Suspicion / Vigilance

PC1 (Valence): -0.82 Negative

PC2 (Disposition): 0.82

Role in Research

This story is one of 1,000 stories generated for the emotion suspicious. During extraction, it was fed through Gemma4-31B and its hidden state activations were captured at 11 layers.

The mean activation across all 1,000 suspicious stories, after denoising with neutral dialogue baselines, produces the suspicious emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the suspicious vector is projected through the unembedding matrix.

Promoted:

`🤔`	0.420
`🤔`	0.415
`🧐`	0.362
`疑`	0.352
`suspicious`	0.352

Suppressed:

`and`	-0.307
`alleviate`	-0.301
`ได้รับการ`	-0.298
`allevi`	-0.285
`mouthful`	-0.284